{
  "id": 132492,
  "title": "list of all mixup variant",
  "url": "/competitions/bengaliai-cv19/discussion/132492",
  "author_name": "",
  "post_date": "2020-02-26T09:32:23.821030100Z",
  "votes": 32,
  "comment_count": 11,
  "views": 0,
  "content": "<p>please feel free to add more:</p>\n\n<p>```</p>\n\n<h1>cutout</h1>\n\n<p>\"Improved Regularization of Convolutional Neural Networks with Cutout\" - Terrance DeVries, arvix 2017\n<a href=\"https://arxiv.org/abs/1708.04552\">https://arxiv.org/abs/1708.04552</a></p>\n\n<h1>mixup</h1>\n\n<p>\"mixup: Beyond Empirical Risk Minimization\" - Hongyi Zhang, arvix 2017\n<a href=\"https://arxiv.org/abs/1710.09412\">https://arxiv.org/abs/1710.09412</a></p>\n\n<h1>manifold mixup</h1>\n\n<p>\"Manifold Mixup: Better Representations by Interpolating Hidden States\" - Vikas Verma, arvix 2018\n<a href=\"https://arxiv.org/abs/1806.05236\">https://arxiv.org/abs/1806.05236</a></p>\n\n<h1>cutmix</h1>\n\n<p>\"CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Features\" - Sangdoo Yun, iccv 2019\n<a href=\"https://arxiv.org/abs/1905.04899\">https://arxiv.org/abs/1905.04899</a></p>\n\n<h1>augmix</h1>\n\n<p>\"AugMix: A Simple Data Processing Method to Improve Robustness and Uncertainty\" - Dan Hendrycks, arvix 2019\n<a href=\"https://arxiv.org/abs/1912.02781\">https://arxiv.org/abs/1912.02781</a></p>\n\n<h1>gridmask</h1>\n\n<p>\"GridMask Data Augmentation\" - Pengguang Chen, arvix 2020\n<a href=\"https://arxiv.org/abs/2001.04086\">https://arxiv.org/abs/2001.04086</a></p>\n\n<h1>dropblock</h1>\n\n<p>\"DropBlock: A regularization method for convolutional networks\" - Golnaz Ghiasi, nips 2019\n<a href=\"https://arxiv.org/abs/1810.12890\">https://arxiv.org/abs/1810.12890</a></p>\n\n<p>```</p>",
  "messages": [
    {
      "id": "756971",
      "postDate": "02/26/2020 09:32:23",
      "content": "<p>please feel free to add more:</p>\n\n<p>```</p>\n\n<h1>cutout</h1>\n\n<p>\"Improved Regularization of Convolutional Neural Networks with Cutout\" - Terrance DeVries, arvix 2017\n<a href=\"https://arxiv.org/abs/1708.04552\">https://arxiv.org/abs/1708.04552</a></p>\n\n<h1>mixup</h1>\n\n<p>\"mixup: Beyond Empirical Risk Minimization\" - Hongyi Zhang, arvix 2017\n<a href=\"https://arxiv.org/abs/1710.09412\">https://arxiv.org/abs/1710.09412</a></p>\n\n<h1>manifold mixup</h1>\n\n<p>\"Manifold Mixup: Better Representations by Interpolating Hidden States\" - Vikas Verma, arvix 2018\n<a href=\"https://arxiv.org/abs/1806.05236\">https://arxiv.org/abs/1806.05236</a></p>\n\n<h1>cutmix</h1>\n\n<p>\"CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Features\" - Sangdoo Yun, iccv 2019\n<a href=\"https://arxiv.org/abs/1905.04899\">https://arxiv.org/abs/1905.04899</a></p>\n\n<h1>augmix</h1>\n\n<p>\"AugMix: A Simple Data Processing Method to Improve Robustness and Uncertainty\" - Dan Hendrycks, arvix 2019\n<a href=\"https://arxiv.org/abs/1912.02781\">https://arxiv.org/abs/1912.02781</a></p>\n\n<h1>gridmask</h1>\n\n<p>\"GridMask Data Augmentation\" - Pengguang Chen, arvix 2020\n<a href=\"https://arxiv.org/abs/2001.04086\">https://arxiv.org/abs/2001.04086</a></p>\n\n<h1>dropblock</h1>\n\n<p>\"DropBlock: A regularization method for convolutional networks\" - Golnaz Ghiasi, nips 2019\n<a href=\"https://arxiv.org/abs/1810.12890\">https://arxiv.org/abs/1810.12890</a></p>\n\n<p>```</p>",
      "rawMarkdown": "please feel free to add more:\n\n```\ncutout\n=======\n\"Improved Regularization of Convolutional Neural Networks with Cutout\" - Terrance DeVries, arvix 2017\nhttps://arxiv.org/abs/1708.04552\n\nmixup\n=======\n\"mixup: Beyond Empirical Risk Minimization\" - Hongyi Zhang, arvix 2017\nhttps://arxiv.org/abs/1710.09412\n\n\nmanifold mixup\n=======\n\"Manifold Mixup: Better Representations by Interpolating Hidden States\" - Vikas Verma, arvix 2018\nhttps://arxiv.org/abs/1806.05236\n\n\ncutmix\n=======\n\"CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Features\" - Sangdoo Yun, iccv 2019\nhttps://arxiv.org/abs/1905.04899\n\n\naugmix\n=======\n\"AugMix: A Simple Data Processing Method to Improve Robustness and Uncertainty\" - Dan Hendrycks, arvix 2019\nhttps://arxiv.org/abs/1912.02781\n\n\ngridmask\n=======\n\"GridMask Data Augmentation\" - Pengguang Chen, arvix 2020\nhttps://arxiv.org/abs/2001.04086\n\n\ndropblock\n=======\n\"DropBlock: A regularization method for convolutional networks\" - Golnaz Ghiasi, nips 2019\nhttps://arxiv.org/abs/1810.12890\n\n```",
      "votes": null
    },
    {
      "id": "756972",
      "postDate": "02/26/2020 09:37:01",
      "content": "<h1>ricap</h1>\n\n<p>\"Data Augmentation using Random Image Cropping and Patching for Deep CNNs\"\n<a href=\"https://arxiv.org/abs/1811.09030\">https://arxiv.org/abs/1811.09030</a></p>",
      "rawMarkdown": "ricap\n=======\n\"Data Augmentation using Random Image Cropping and Patching for Deep CNNs\"\n[https://arxiv.org/abs/1811.09030](https://arxiv.org/abs/1811.09030)",
      "votes": null
    },
    {
      "id": "756974",
      "postDate": "02/26/2020 09:39:10",
      "content": "<p>\"Random Erasing Data Augmentation\" - Zhun Zhong, arvix 2017\n<a href=\"https://arxiv.org/abs/1708.04896\">https://arxiv.org/abs/1708.04896</a></p>",
      "rawMarkdown": "\"Random Erasing Data Augmentation\" - Zhun Zhong, arvix 2017\nhttps://arxiv.org/abs/1708.04896",
      "votes": null
    },
    {
      "id": "756977",
      "postDate": "02/26/2020 09:41:49",
      "content": "<p>Progressive sprinkles\n<a href=\"https://medium.com/@lessw/progressive-sprinkles-a-new-data-augmentation-for-cnns-and-helps-achieve-new-98-nih-malaria-6056965f671a\">https://medium.com/@lessw/progressive-sprinkles-a-new-data-augmentation-for-cnns-and-helps-achieve-new-98-nih-malaria-6056965f671a</a></p>",
      "rawMarkdown": "Progressive sprinkles\nhttps://medium.com/@lessw/progressive-sprinkles-a-new-data-augmentation-for-cnns-and-helps-achieve-new-98-nih-malaria-6056965f671a",
      "votes": null
    },
    {
      "id": "757287",
      "postDate": "02/26/2020 15:57:29",
      "content": "<p>not quite like mixup, but related is a set of noise regularization methods (or noise augmentation) related to dropout, drop channel, shake-shake, shake-drop ...</p>",
      "rawMarkdown": "not quite like mixup, but related is a set of noise regularization methods (or noise augmentation) related to dropout, drop channel, shake-shake, shake-drop ...",
      "votes": null
    },
    {
      "id": "761051",
      "postDate": "03/02/2020 04:35:09",
      "content": "<p>FMix: FMix improves performance over MixUp and CutMix for a number of state-of-the- art models across a range of data sets and problem settings</p>\n\n<p><a href=\"https://arxiv.xilesou.top/pdf/2002.12047.pdf\">https://arxiv.xilesou.top/pdf/2002.12047.pdf</a></p>",
      "rawMarkdown": "FMix: FMix improves performance over MixUp and CutMix for a number of state-of-the- art models across a range of data sets and problem settings\n \n \nhttps://arxiv.xilesou.top/pdf/2002.12047.pdf",
      "votes": null
    },
    {
      "id": "761053",
      "postDate": "03/02/2020 04:37:09",
      "content": "<p>MaxUp: A Simple Way to Improve Generalization of Neural Network Training\nC Gong, T Ren, M Ye, Q Liu - arXiv preprint arXiv:2002.09024, 2020 - arxiv.org</p>",
      "rawMarkdown": "MaxUp: A Simple Way to Improve Generalization of Neural Network Training\nC Gong, T Ren, M Ye, Q Liu - arXiv preprint arXiv:2002.09024, 2020 - arxiv.org",
      "votes": null
    },
    {
      "id": "762355",
      "postDate": "03/03/2020 13:10:00",
      "content": "<p>Any implemntation of dropblock for Keras/Tensorflow 2 backend  ?</p>",
      "rawMarkdown": "Any implemntation of dropblock for Keras/Tensorflow 2 backend  ?",
      "votes": null
    },
    {
      "id": "762364",
      "postDate": "03/03/2020 13:23:04",
      "content": "<p><a href=\"https://github.com/iantimmis/DropBlock-Keras-Implementation\">https://github.com/iantimmis/DropBlock-Keras-Implementation</a>\n<a href=\"https://github.com/CyberZHG/keras-drop-block\">https://github.com/CyberZHG/keras-drop-block</a></p>\n\n<p>I didn't try it but will. </p>",
      "rawMarkdown": "https://github.com/iantimmis/DropBlock-Keras-Implementation\nhttps://github.com/CyberZHG/keras-drop-block\n\nI didn't try it but will.",
      "votes": null
    },
    {
      "id": "762736",
      "postDate": "03/03/2020 19:14:09",
      "content": "<p><a href=\"/ipythonx\">@ipythonx</a> I tried it earlier but didn't work well with tf2 keras</p>",
      "rawMarkdown": "ipythonx I tried it earlier but didn't work well with tf2 keras",
      "votes": null
    },
    {
      "id": "762756",
      "postDate": "03/03/2020 19:34:44",
      "content": "<p>I always wonder how much time do you spend on this competition? Your experiment and sharing materials make you seem like a whole team to me!😃 That's really impressive!</p>",
      "rawMarkdown": "I always wonder how much time do you spend on this competition? Your experiment and sharing materials make you seem like a whole team to me!😃 That's really impressive!",
      "votes": null
    },
    {
      "id": "764660",
      "postDate": "03/05/2020 17:47:16",
      "content": "<p>batchboost: regularization for stabilizing training with resistance to underfitting &amp; overfitting\nMaciej A. Czyzewski</p>\n\n<p><img src=\"https://github.com/maciejczyzewski/batchboost/raw/master/figures/figure-feeding.png\" alt=\"\">\n<img src=\"https://github.com/maciejczyzewski/batchboost/raw/master/figures/figure-abstract.png\" alt=\"\"></p>",
      "rawMarkdown": "batchboost: regularization for stabilizing training with resistance to underfitting &amp; overfitting\nMaciej A. Czyzewski\n\n\n![](https://github.com/maciejczyzewski/batchboost/raw/master/figures/figure-feeding.png)\n![](https://github.com/maciejczyzewski/batchboost/raw/master/figures/figure-abstract.png)",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 756972,
      "author_name": "pestipeti",
      "author_url": "",
      "post_date": "02/26/2020 09:37:01",
      "content": "<h1>ricap</h1>\n\n<p>\"Data Augmentation using Random Image Cropping and Patching for Deep CNNs\"\n<a href=\"https://arxiv.org/abs/1811.09030\">https://arxiv.org/abs/1811.09030</a></p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 756974,
      "author_name": "hengck23",
      "author_url": "",
      "post_date": "02/26/2020 09:39:10",
      "content": "<p>\"Random Erasing Data Augmentation\" - Zhun Zhong, arvix 2017\n<a href=\"https://arxiv.org/abs/1708.04896\">https://arxiv.org/abs/1708.04896</a></p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 756977,
      "author_name": "andreyzotov",
      "author_url": "",
      "post_date": "02/26/2020 09:41:49",
      "content": "<p>Progressive sprinkles\n<a href=\"https://medium.com/@lessw/progressive-sprinkles-a-new-data-augmentation-for-cnns-and-helps-achieve-new-98-nih-malaria-6056965f671a\">https://medium.com/@lessw/progressive-sprinkles-a-new-data-augmentation-for-cnns-and-helps-achieve-new-98-nih-malaria-6056965f671a</a></p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 757287,
      "author_name": "hengck23",
      "author_url": "",
      "post_date": "02/26/2020 15:57:29",
      "content": "<p>not quite like mixup, but related is a set of noise regularization methods (or noise augmentation) related to dropout, drop channel, shake-shake, shake-drop ...</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 761051,
      "author_name": "hengck23",
      "author_url": "",
      "post_date": "03/02/2020 04:35:09",
      "content": "<p>FMix: FMix improves performance over MixUp and CutMix for a number of state-of-the- art models across a range of data sets and problem settings</p>\n\n<p><a href=\"https://arxiv.xilesou.top/pdf/2002.12047.pdf\">https://arxiv.xilesou.top/pdf/2002.12047.pdf</a></p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 761053,
      "author_name": "hengck23",
      "author_url": "",
      "post_date": "03/02/2020 04:37:09",
      "content": "<p>MaxUp: A Simple Way to Improve Generalization of Neural Network Training\nC Gong, T Ren, M Ye, Q Liu - arXiv preprint arXiv:2002.09024, 2020 - arxiv.org</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 762355,
      "author_name": "ma7555",
      "author_url": "",
      "post_date": "03/03/2020 13:10:00",
      "content": "<p>Any implemntation of dropblock for Keras/Tensorflow 2 backend  ?</p>",
      "votes": null,
      "replies": [
        {
          "id": 762364,
          "author_name": "ipythonx",
          "author_url": "",
          "post_date": "03/03/2020 13:23:04",
          "content": "<p><a href=\"https://github.com/iantimmis/DropBlock-Keras-Implementation\">https://github.com/iantimmis/DropBlock-Keras-Implementation</a>\n<a href=\"https://github.com/CyberZHG/keras-drop-block\">https://github.com/CyberZHG/keras-drop-block</a></p>\n\n<p>I didn't try it but will. </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 762736,
          "author_name": "ma7555",
          "author_url": "",
          "post_date": "03/03/2020 19:14:09",
          "content": "<p><a href=\"/ipythonx\">@ipythonx</a> I tried it earlier but didn't work well with tf2 keras</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 762756,
      "author_name": "yuanlin08",
      "author_url": "",
      "post_date": "03/03/2020 19:34:44",
      "content": "<p>I always wonder how much time do you spend on this competition? Your experiment and sharing materials make you seem like a whole team to me!😃 That's really impressive!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 764660,
      "author_name": "hengck23",
      "author_url": "",
      "post_date": "03/05/2020 17:47:16",
      "content": "<p>batchboost: regularization for stabilizing training with resistance to underfitting &amp; overfitting\nMaciej A. Czyzewski</p>\n\n<p><img src=\"https://github.com/maciejczyzewski/batchboost/raw/master/figures/figure-feeding.png\" alt=\"\">\n<img src=\"https://github.com/maciejczyzewski/batchboost/raw/master/figures/figure-abstract.png\" alt=\"\"></p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "756971": "please feel free to add more:\n\n```\ncutout\n=======\n\"Improved Regularization of Convolutional Neural Networks with Cutout\" - Terrance DeVries, arvix 2017\nhttps://arxiv.org/abs/1708.04552\n\nmixup\n=======\n\"mixup: Beyond Empirical Risk Minimization\" - Hongyi Zhang, arvix 2017\nhttps://arxiv.org/abs/1710.09412\n\n\nmanifold mixup\n=======\n\"Manifold Mixup: Better Representations by Interpolating Hidden States\" - Vikas Verma, arvix 2018\nhttps://arxiv.org/abs/1806.05236\n\n\ncutmix\n=======\n\"CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Features\" - Sangdoo Yun, iccv 2019\nhttps://arxiv.org/abs/1905.04899\n\n\naugmix\n=======\n\"AugMix: A Simple Data Processing Method to Improve Robustness and Uncertainty\" - Dan Hendrycks, arvix 2019\nhttps://arxiv.org/abs/1912.02781\n\n\ngridmask\n=======\n\"GridMask Data Augmentation\" - Pengguang Chen, arvix 2020\nhttps://arxiv.org/abs/2001.04086\n\n\ndropblock\n=======\n\"DropBlock: A regularization method for convolutional networks\" - Golnaz Ghiasi, nips 2019\nhttps://arxiv.org/abs/1810.12890\n\n```",
    "756972": "ricap\n=======\n\"Data Augmentation using Random Image Cropping and Patching for Deep CNNs\"\n[https://arxiv.org/abs/1811.09030](https://arxiv.org/abs/1811.09030)",
    "756974": "\"Random Erasing Data Augmentation\" - Zhun Zhong, arvix 2017\nhttps://arxiv.org/abs/1708.04896",
    "756977": "Progressive sprinkles\nhttps://medium.com/@lessw/progressive-sprinkles-a-new-data-augmentation-for-cnns-and-helps-achieve-new-98-nih-malaria-6056965f671a",
    "757287": "not quite like mixup, but related is a set of noise regularization methods (or noise augmentation) related to dropout, drop channel, shake-shake, shake-drop ...",
    "761051": "FMix: FMix improves performance over MixUp and CutMix for a number of state-of-the- art models across a range of data sets and problem settings\n \n \nhttps://arxiv.xilesou.top/pdf/2002.12047.pdf",
    "761053": "MaxUp: A Simple Way to Improve Generalization of Neural Network Training\nC Gong, T Ren, M Ye, Q Liu - arXiv preprint arXiv:2002.09024, 2020 - arxiv.org",
    "762355": "Any implemntation of dropblock for Keras/Tensorflow 2 backend  ?",
    "762364": "https://github.com/iantimmis/DropBlock-Keras-Implementation\nhttps://github.com/CyberZHG/keras-drop-block\n\nI didn't try it but will.",
    "762736": "ipythonx I tried it earlier but didn't work well with tf2 keras",
    "762756": "I always wonder how much time do you spend on this competition? Your experiment and sharing materials make you seem like a whole team to me!😃 That's really impressive!",
    "764660": "batchboost: regularization for stabilizing training with resistance to underfitting &amp; overfitting\nMaciej A. Czyzewski\n\n\n![](https://github.com/maciejczyzewski/batchboost/raw/master/figures/figure-feeding.png)\n![](https://github.com/maciejczyzewski/batchboost/raw/master/figures/figure-abstract.png)"
  },
  "source": "meta"
}